Wristwatch daily summaries, 1 Jan – 31 Dec 2025 — the year before the 2026 companion chart. The score (allDayStress) is HRV-based, not self-reported: the wristwatch reads heart-rate variability from the wrist sensor every few minutes it can get a clean signal, and scores 0–100 by how much that pattern reflects sympathetic (low HRV variability, tense) vs. parasympathetic (higher variability, relaxed) autonomic balance — the device's own bands are Rest 0–25, Low 26–50, Medium 51–75, High 76–100. Travel bands below only start 23 Aug — that is as far back as Uwe's Travel Schedule goes; the score and step data for 1 Jan – 22 Aug is real and complete, just without a confirmed place attached, so it renders unbanded rather than mislabelled as Bansko. Travel bands are shown only where Uwe confirmed the dates directly (via his own Travel Schedule) or a GPS-tagged activity confirms a specific day — no dates are inferred from the score or step pattern alone.
1. Average daily score (HRV-based, 0–100)
2. Daily steps
3. Travel
Reading this: gaps in the two data charts mean no wearable data that day (watch not synced or not worn) — 39 such days across the year. A raw value of −1 ("not calculated") is excluded, not plotted as zero. Travel evidence standard (the same standard used on the 2026 chart): a date only appears as a confirmed travel block when Uwe stated it directly (via his Travel Schedule) or a GPS-tagged wristwatch activity names the place. Nothing here is guessed from the score or step shape alone — that includes the blank first two-thirds of the year: 1 Jan – 22 Aug has real score and step data (charted above) but no confirmed location, since the Travel Schedule only starts 23 Aug, so it is left unbanded rather than assumed to be Bansko. Within the travel-confirmed era, the closest thing to a Bansko baseline is the handful of schedule-confirmed home days between trips (20–24 Sep, 12–17 Oct): score 14–60, averaging 32.2 across just 11 days — too small and noisy a sample to anchor the comparisons below, so the whole-year average (24.8) stands in as the reference line instead. See "Patterns across 2025" for the detail. Also visible without any travel claim attached: December was the tensest month of the year (32.8) and April the calmest (18.6) — but April sits entirely inside the location-unconfirmed Jan–Aug stretch, so nothing about place can be read into it. The Vietnam leg starting 16 Dec continues past 31 Dec into 2026 — see the companion chart for the rest of that stay (through 11 Mar) and everything after.
4. An open question: does who matters more than where?
2025's calmest identifiable stretches share a similar shape to 2026's: relaxed, low-logistics company rather than an easier destination. Uzès, four days with a friend in the middle of the Austria/Switzerland/France trip, is the calmest multi-day stretch in the whole travel-confirmed year (23.2) — calmer even than the sailing week in Turkey (28.2) that became 2025's most obviously "relaxing" trip on paper. Kyoto, six days with a friend partway through Japan (26.3), is the next calmest. Both were quiet, one-on-one stays with nothing to organise once there — not the destinations most people would nominate as 2025's highlight, but the readings that came closest to home.
The tensest stretches sit at the opposite end of the same trips: the Nomad Explore conference days in Chiayi (37.3) and the days with a friend in Taichung that followed it (37.8) — both inside the busiest, most event- and people-dense week of the whole year, on a trip that also carried the highest step count of 2025 (9.4k/day). Vietnam's December stay, arriving into a new long-term base with a friend, ran tenser again (32.8) than anywhere earlier in the year — though the 2026 chart shows that same stay settling to 29.4 once it becomes routine. None of this is as clean a test as 2026's: every 2025 sample here is small (n=3–16), and two-thirds of the year has no confirmed location to compare against at all. But the shape recurs — quiet, one-on-one time reads calmer than busy, event-driven company, independent of the destination itself.
5. How much does a point actually mean? — e.g. Turkey/sailing 28.9 vs. whole-year 24.8
As with the 2026 chart, there's no official figure for how many points separate a real change from ordinary noise — the closest evidence is a 28-day tracking study reporting a day-to-day spread of roughly 15 points for one person†, so a 1–2 point gap between two single days means nothing on its own. Averages behave differently: noise shrinks with the square root of the day count, so each 2025 trip's error bar runs roughly ±2.7 to ±3.9 points depending on its length, against roughly ±0.8 for the 326-day whole-year average. Four of the five confirmed trips clear that combined bar by a moderate margin — Austria/CH/FR (30.0, ~1.5×), Japan (30.3, ~1.5×), Taiwan (32.2, ~1.8×) and Vietnam's December stay (32.8, ~1.75×) — while Turkey/sailing, the calmest of the five on paper (28.9), does not (~0.9×): its gap against the whole-year average is smaller than its own noise band, so calling it 2025's "calmest trip" is a real but modest claim, not a strong one. The clearest 2025 gap by far is the monthly one — December (32.8, n=31) against April (18.6, n=25) is a 14.2-point gap against a combined error bar of roughly ±5.7, about 2.5× — a real difference, though April sits entirely inside the location-unconfirmed stretch, so it cannot be attributed to any place.
Every number here also comes from the same wristwatch and the same on-device algorithm as the 2026 chart, used consistently — a consistently-biased ruler still measures differences correctly, so the comparisons above hold even if the absolute score runs 5–10 points off a lab instrument (independent validation found it correlates only moderately, r≈−0.6, with lab-grade HRV‡). Read every number on this page as directional, not exact.
† Wristwatch stress-score day-to-day variance study (Authorea preprint) — "variance" here may be a loose spread rather than a strict statistical variance; this box uses the larger, more conservative reading either way.
‡ Independent validation study, bioRxiv 2025.
6. Patterns across 2025
Average score by confirmed stretch, lower is calmer, against the 2025 whole-year average (dashed line) — unlike 2026, the home stretch (Bansko) is too small and noisy a sample here (11 days, range 14–60) to serve as a reliable baseline, so the whole-year figure stands in instead, and is shown as its own bar too.
Colour marks a real relationship found in the data; this year the null result is rest minutes, not steps — the opposite of 2026.
7. What the score measures
The wristwatch's daily figure is built on Firstbeat Analytics' heart-rate-variability (HRV) methodology: the wrist sensor reads the time between heartbeats every few minutes it gets a clean signal, and the pattern in that variability is scored 0–100 by how far it leans toward sympathetic dominance (tense, low variability) versus parasympathetic dominance (calm, higher variability) — respiration and movement data are used only to filter out exertion noise, not to raise or lower the score directly.1 An independent 2025 validation study found the score correlated significantly with lab-grade HRV measures (RMSSD, SD2/SD1) and with heart rate itself against instrumented references.2
Two things are worth keeping in mind when reading the sections above. First, HRV-based readings and how someone actually feels do not always move together — reviews of the wider literature treat the two as related but distinct signals3, worth remembering wherever the same-day score does not match what the itinerary alone would suggest (the Chiayi conference days, or the quiet stay in Uzès, for instance). Second, travel itself pulls in two directions in the research: unstructured getaways are associated with measurably lower cortisol and better HRV on average across studies4,5, while long-haul travel and disrupted routine push the other way, independently linked to reduced HRV regardless of subjective experience6. Both effects show up in 2025 too — quiet, one-on-one stays (Uzès, Kyoto) calmer than the noisy small Bansko sample, event-heavy stretches (Chiayi, Taichung) tenser than the year's average — though with only a third of the year carrying a confirmed location and every sample smaller than 2026's, the pattern here is more suggestive than conclusive.
8. Travel index
| Place | Dates | HRV index (avg) | Steps (avg) | Notes | |
|---|---|---|---|---|---|
| Austria / Switzerland / France | 25 Aug – 19 Sep, 18–25 Oct | 30.0 | 5.3k | Road trip: Bansko → Feldkirch (Austria) → Swiss Coliving (Liddes/Grimentz) → Zermatt with a friend → Uzès, France with a friend → a group party near Vézac (12–14 Sep) → Zürich stopover with friends → back to Bansko 20 Sep. Second leg 18–25 Oct: Sofia → Munich → Feldkirch (appointment, admin). | |
| Turkey | 25 Sep – 11 Oct | 28.9 | 5.3k | Drive Bansko → Marmaris, then a 2-week sailing trip (27 Sep – 11 Oct) with a companion. GPS-confirmed: 3 Oct (Marmaris), 6 & 9 Oct (Mugla). | |
| Japan | 26 Oct – 24 Nov | 30.3 | 7.5k | Tokyo (2 days) → Hakuba (speaking event, 31 Oct – 2 Nov; extended stay) → Kanazawa → Kyoto with a friend (7 days) → Okinawa island retreat with a friend (2 weeks). GPS-confirmed: 3 Nov (Hakuba Village). | |
| Taiwan | 25 Nov – 15 Dec | 32.2 | 9.4k | Taipei (alone) → Hengchun with friends → Kaohsiung (alone) → Chiayi (Nomad Explore 2025 conference) → Taichung with a friend → Taipei. GPS-confirmed: 6 Dec (Zhuqi Township, near Chiayi). | |
| Vietnam | 16 – 31 Dec (continues into 2026) | 32.8 | 5.1k | Flight Taipei → Da Nang, with a friend, into the long stay that runs through 11 Mar 2026 — see the 2026 chart for the rest. |
9. Activity comparison
Only 5 GPS-tracked activities exist in the whole 2025 wristwatch export — 3 stand-up-paddleboarding sessions during the Turkey trip, plus one walk each in Hakuba and Zhuqi. Sailing itself was never started as a tracked activity (no GPS session for time underway), so there's no per-activity stress metric for it — the wristwatch only attaches the HRV score to whole days, not activity segments. Most of the comparison below is therefore day-level: the average score across named periods pulled from the Travel Schedule notes — a friend's visit, a conference, an island retreat — set against the rest of each trip with no specific note attached. This section only covers the five location-confirmed trips; the location-unconfirmed Jan–22 Aug stretch has no named periods to compare. Bar shade marks how each row is evidenced: solid = a logged wristwatch activity, medium = a named period from the Travel Schedule, light = the rest of that trip (or, for Vietnam's December stay, the whole thing — it has no sub-period breakdown yet).
Answer: the two calmest named periods anywhere in the confirmed months were quiet, one-on-one stays — Uzès with a friend (23.2) and Kyoto with a friend (26.3) — while the tensest were the most event- and people-dense: the Chiayi conference days (37.3) and the Taichung stay with a friend that followed it (37.8), both inside Taiwan's busiest stretch. The three GPS-tracked activities sit scattered across the middle of the range rather than clustering low — SUP paddleboarding (27.7), the Hakuba walk (29.0) and the Zhuqi walk (37.0) — echoing 2026's own finding that deliberate, planned exertion does not read as calmer than its surrounding trip. Every row here is a small sample (n=1–24); read as suggestive, not conclusive, same as the 2026 chart.